[ https://issues.jboss.org/browse/JGRP-2253?page=com.atlassian.jira.plugin.... ] Sibin Karnavar commented on JGRP-2253: -------------------------------------- <<One thing you should do is to separate TCP.bind_port (7803), TCP.diagnostics_port (7806) and FD_SOCK.start_port (7804)>> Currently we have defined my bind port as 7803 and port_range as 0. So There is no chance that a member can use 7804 as my TCP bind port. We did this because we want to make sure only one port is dedicated for TCP bind port. In addition to that, we have only one member running in one EC2 instance. Do you think it is still an issue? if So What is the other best way to separate it out. <<Note that I've run JGroups on AWS before, and I didn't define any start_port or client_bind_port in FD_SOCK, so these ports are picked by the OS.>> We have defined a security group inbound ports in AWS so only certain port is allows to access an EC2 instance. If JGroup uses a random client_bind_port, then it won't work. That's why we have mentioned a specific port. I was able to make FD_SOCK work after defining that port. Thats why kill -9 always work. But for some reason terminating EC2 instance from console doesn't work well. -- This message was sent by Atlassian JIRA (v7.5.0#75005)